Serverless Frameworkで AWS Lambda Layersを作る

目的

Serverless Frameworkを使って、Lambda関数と同時にLambda Layerを作成する。

ディレクトリ構成

sls-example
  ├── handle.py
  ├── layers
  │   └── pillow
  │       ├── package.sh
  │       └── requirements.txt
  └── serverless.yml

requirements.txt

pillow==6.0.0

package.sh

#!/bin/sh

export OUTPUT_DIR="python"
rm -rf ${OUTPUT_DIR} && mkdir -p ${OUTPUT_DIR}
docker run --rm -v $(pwd):/var/task -w /var/task lambci/lambda:build-python3.6 \
    pip install -r requirements.txt -t ${OUTPUT_DIR} 

serverless.yml

service: sls-example

provider:
  name: aws
  runtime: python3.6
  stage: dev
  region: us-west-2
plugins:
  - serverless-python-requirements
package:
  individually: true

# AWS Lambda Layers
layers:
  pillow:
    path: layers/pillow
    name: ${self:service}-pillow
    compatibleRuntimes:
      - python3.6
    allowedAccounts:
      - '*'

# AWS Lambda Functions
functions:
  analyze:
    handler: analysis_lambda.analyze
    layers:
      - {Ref: PillowLambdaLayer}
#     - arn:aws:lambda:${self:provider.region}:xxxxxxxxxxx:layer:{service名}-{layer名}:{バージョン}

デプロイ

cd pillow
./package.sh
cd sls-example
sls deploy